RFR: 8319315: Sending a SysexMessage starting with 0xF7 leads to JVM crash

Sergey Bylokhov serb at openjdk.org
Mon Nov 6 18:48:40 UTC 2023


On Thu, 2 Nov 2023 13:55:41 GMT, Alec Su <duke at openjdk.org> wrote:

> The bug on Windows doesn't generate any error messages. During the testcase run, every "Open MIDI port" message should be followed by an "All SysexMessage sent" message. Alternatively, you can check `%ERRORLEVEL%` to determine if there's an error when JVM exits.
> 
> On Linux, there are no software MIDI devices provided by default. To reproduce this bug on Linux, you need to connect a real MIDI device to Linux or create a virtual MIDI device by running the command `sudo modprobe snd_virmidi midi_devs=1`.
> 
> The bug occurred because a pointer was increased on Line 137, then it attempted to release the modified pointer on Line 143.
> 
> https://github.com/openjdk/jdk/blob/e9d19d0fffc47119d0d4f756833ec87cd0a6331e/src/java.desktop/share/native/libjsound/MidiOutDevice.c#L128-L144
> 
> To fix this, I added a variable `msg` to store the modified pointer to prevent `data` pointer being modified.

It seems the bug is a duplicate of the next one https://bugs.openjdk.org/browse/JDK-8301310 ?

-------------

PR Comment: https://git.openjdk.org/jdk/pull/16477#issuecomment-1795911706


More information about the client-libs-dev mailing list